wide-bandgap semiconductors in space: appreciating the

National Aeronautics and Space Administration

Wide-Bandgap Semiconductors in Space: Appreciating the Benefits but Understanding the Risks

Jean-Marie LauensteinNASA GSFC, Greenbelt, MD, USA

1To be presented by J.-M. Lauenstein at the Radiation Effects on Components and Systems (RADECS) Topical Day, Gothenburg, Sweden, September 16, 2018

Acronyms and Abbreviations


Acronym Definition

2-D Two Dimensional

AlGaN/AlN Aluminum Gallium Nitride/Aluminum Nitride

BVDSS Drain-Source Breakdown Voltage

C Carbon

DDD Displacement Damage Dose

EC Conduction Band Energy

EF Fermi Level Energy

Egap Bandgap Energy

ESA European Space Agency

EV Valance Band Energy

FIT Failures In Time

FOM Figure of Merit

FY Fiscal Year

GaAs Gallium Arsenide

GaN Gallium Nitride

Ga2O3 Gallium Oxide

GCR Galactic Cosmic Ray

GEO Geostationary Earth Orbit

HEMT High Electron Mobility Transistor

Acronym Definition

ID Drain Current

IDSS Drain-Source Leakage Current

IG Gate Current

IGSS Gate-Source Leakage Current

InP Indium Phosphide

IR Reverse Current

ISS International Space Station

JBS Junction Barrier Schottky diode

JFET Junction Field Effect Transistor

LEO Low Earth Orbit

LET Linear Energy Transfer

MESFET Metal-Semiconductor Field Effect Transistor

MOSFET Metal Oxide Semiconductor Field Effect Transistor

NO Nitric Oxide

PIGS Post-Irradiation Gate Stress

R&D Research and Development

RDS_ON On-State Drain-Source Resistance

RF Radio Frequency

RHA Radiation Hardness Assurance

RHBP Radiation Hardened By Process

Acronym Definition

SBD Schottky Barrier Diode

SBIR Small Business Innovative Research

SEB/GR Single-Event Burnout/Gate Rupture

SEE Single-Event Effect

Si Silicon

SiC Silicon Carbide

Sn Tin

SOA State Of the Art; Safe Operating Area

STMD Space Technology Mission Directorate

SWAP Size, Weight, And Power

TAMU Texas A&M University cyclotron facility

TID Total Ionizing Dose

UWBG Ultra-Wide Bandgap

VDMOS Vertical Double-diffused MOSFET

VDS Drain-Source Voltage

VGS Gate-Source Voltage

VR Reverse-bias Voltage

VTH Gate Threshold Voltage

WBG Wide Bandgap

Xe Xenon

•

Outline


Overview

WBG

RH

AThe Future

Reprinted with permission from Yao, et al., JVSTB, vol. 35, p. 03D113. Copyright 2017, American Vacuum Society

Sn-doped Ga2O3 wafer

• Overview– Bandgaps: the simple explanation– Wide-bandgap (WBG) technology advantages & applications– Space environment hazards

• Radiation Hardness Assurance (RHA) of Commercially-Mature WBG Semiconductors – SiC discrete power devices– GaN high electron mobility transistors (HEMTs):

• RF• Enhancement-mode

• A Glimpse into the WBG Semiconductor Horizon– Qualification standardization & technology maturation– The future of SiC and GaN– Ultra-Wide Bandgap (UWBG) semiconductors

• Diamond• Ga2O3

0 < ct)

< ro ~

•

( ac) ) I

Wide-Bandgap Technology Overview:What’s a Bandgap? (Simply Speaking)


Conduction Band

ValenceBandE

nerg

y (E

) [eV

]

Egap

• As atoms come together to form a crystal, their electron shells form bands: – the highest-energy band containing electrons at 0 K = Valence Band– the next highest energy band = Conduction Band

• Electrons in the conduction band (and subsequent holes in the valence band) are able to move freely about the lattice and thus conduct current

• In order to become conducting, an electron must have enough energy to jump the gap– this jump can be assisted by traps, or energy states, located in the band gap

Wide bandgap materials have different electrical properties than conventional bandgap materials (e.g., Si) due to their different band structure and band gap

Overview

0 < CD

< ro :E •

Wide-Bandgap Technology Overview:Properties Associated with the Bandgap


• WBG semiconductors can:– Operate at higher temperatures without “going intrinsic”– Block higher voltages for a given thickness of material (breakdown voltage ∝ (Egap)4)– Switch at higher speeds due in part to higher saturation velocities– … and other WBG-material specific advantages

Overview

• 0 7 < ct)

< 6 ro ~

~ Conventional 6 [::=J WBG -

~ Ultra-WBG 5.5

>5 Cl) ...... - 4.8 >-El 4 Cl) C: w

-

3.3 3.4 Q. 3 n, -C)

"O C: n, 2 co

-

1.12 1.4

1 -

0 I I I I I I I

Si GaAs 4H-SiC GaN J3-Ga2O3 C AIGaN/AIN

A Closer Look: SiC vs. Si


SiC out-performs Si on 5 different parameters, lending itself to high-power, high-temperature, and fast-switching applications

Overview

0 < CD

< co :E

Electron Saturation Velocity

[x107 cm/s]

Fast Switching Applications

Melting Point [x1000 °C]

-- - ----------------,

High Temperature Applications

Electric Field [MV/cm]

E si 1- 4H-SiC

Thermal Conductivity [W/cm/°C]

•

A Closer Look: GaN vs. SiC vs. Si


To date, GaN’s upper limit on voltage rating is dictated primarily by device degradation/reliability issues

Overview

0 < CD

< ro ~


[x107 cm/s]


Melting Point

[x1000 °C]

Bandgap Energy ~-[eV] ~ '9~~


~~- 0/4 ~ C"~

' '

Q>~. '9& 0~ ~


- Si - 4H-SiC - GaN_j

Thermal Conductivity

[W/cm/°C]

•

A Closer Look: RF Performance

• RF power amplifiers are critical for radar and communications systems– Choice of semiconductor material will

depend in part on power and frequency needs

• Plot on left is intended only to provide a general idea of the capabilities of each material


DiamondGa2O3

GaN

Out

put P

ower

Frequency

SiC

InPGaAs Si

Future RF amplifiers may combine different semiconductor technologiesfor system optimization

Overview

0 < (1)

< ro :E

L.. Q)

~ 0

Cl.

•

Frequency

• Reduced current for the same power– Smaller-gauge harnesses/cabling

• Decreased design complexity – Reduced need for device stacking

• Smaller, lighter passive circuit elements

• Smaller, lighter cooling elements


Higher Blocking Voltage: Faster Switching Speeds & Lower Losses:

Toyota estimates 80% volume reduction of Prius power control unit using SiC vs. Si

https://newsroom.toyota.co.jp/en/detail/2656842 (Image used with permission)

NASA Solar Electric Propulsion:2.7 tons mass savings by moving to higher voltage power management & distribution

Source: Mercer, AIAA 2011-7252; Image: NASA

Wide-Bandgap Technology Overview:Translation of Properties into Performance

High Power Density:• 5x – 10x smaller footprint for

RF power GaN vs. GaAs• Easier impedance matching

– lower circuit losses

Ka-band power amplifier: 72% area reduction & 4x power increase using GaN vs. GaAsImage modified from: Kong, K. et al., IEEE CSICS, 2014

Overview

•

Power module

Wide-Bandgap Technology Overview:The Space Radiation Environment

• Trapped radiation belts– High & variable flux of protons and

electrons– Concerns include:

• Total ionizing dose (TID)• Displacement damage dose (DDD)• Proton-induced single-event effects (SEE)

• Solar Particle Events (SPE)– Intermittent high flux of moderate-

energy protons, electrons, and ions with atomic number Z = 1 to ~26

– Concerns include TID, DDD, and SEE• Galactic Cosmic Rays (GCR)

– Low flux of very high energy ions of all Z

– Primary concern = SEE


TID, DDD, SEESEE

TIDDDD

Image modified from:

Overview

0 < Cl)

< ro :E

•

Wide-Bandgap Technology Overview:“But I Heard that WBG Semiconductors are Rad-Hard!”

• “Inherent” radiation hardness of WBG semiconductors typically refers to their tolerance of total dose:– Both the ionization energy and threshold energy for defect formation (atomic bond strength)

exceed that for Si• Must also consider the material density (interaction cross section)

– Early WBG devices did not have gate oxides– For the same working voltage range, SiC and other doped WBG materials can have a higher

doping concentration to decrease sensitivity to carrier removal– Operation of WBG devices at high temperature results in greater mobility of defects, reducing

formation of defect complexes


Overview

0 < CD

< c;;· ~

•

SILICON CARBIDE POWER DEVICE RADIATION EFFECTS


•

SiC Outline

• Effects of radiation:– Displacement damage dose

• Diodes, JFETs

– Total ionizing dose• MOSFETs

– Single-event effects• Diodes, MOSFETs

• Radiation hardness assurance (RHA) guidance for risk-tolerant applications

• RHA considerations: looking forward


4H-SiC

6H-SiC

Davidsson, New J. Phys. 20 (2018) 023035Used with permission.

WBG

RH

A

~ co G)

::0 I )>

B,k

C,h

B, k

A,h

0 1

• B, k2

C

C, kl

A,h

C, k2

B, kl

A,h

Silicon Carbide:Displacement Damage Dose Effects

• SiC Schottky diodes are DDD-robust– Most mission DDD(Si) levels are << 1010 MeV/g – Europa Clipper is one exception at

1.7∙1010 MeV/g*• (Green dashed vertical line on plot)


Modified from: Harris, IEEE REDW, 2007

*Europa Clipper DDD(Si) behind 100 mil AlFrom: JPL D-80302, Aug 18, 2016.

SiC Schottky Diode Performance with Dose

WBG

RH

A

~ OJ G)

::a I )>

3.0

...... ~ 2.5 <(

1ij 2.0 Cl) C) n, ~ 1.5 0 > 'E 1.0

~ ~ 0.5

-

-

-

-

-

-

0.0 1010

• I I --•- SiC: CSD10030A •

-•-SiC: CSD10120A -

-

• I -• • •• I -I • •-•-•-•·•· • I / • . • •-•-· •

I -I I I

I

1011 1012

Displacement Damage Dose [MeV/g]


• SiC Schottky diodes are DDD-robust• Schottky barrier diodes (SBD) may be

more tolerant than junction barrier diodes (JBS)– Most commercial SiC Schottky’s are JBS

• Protects the Schottky barrier from the high field achievable with WBG materials


Luo, IEEE TNS, 2004

SiC SBD vs. JBS

WBG

RH

A

~ CD G)

:::0 I )>

,...., N

I

E 0 I

<( ...... ~ ti) C: (1)

C ..., C: (1) ... ... ::::, u

102

10°

10·2

104

10-6

10-8

1 o -1 0

-1 0

• ---· ----

SBD pre •·-0-·· SBD post • - - JBS pre

ODD= 1.9x1011 MeV/g - •- JBS post

1 2 3 4 5 6 7 Forward Voltage [V]


• SiC Schottky diodes are DDD-robust• Schottky barrier diodes (SBD) may be

harder than junction barrier diodes (JBS)• Direct comparison with Si is difficult

– Harris, 2007 showed carrier removal rate in SiCSchottky diodes was higher than in Si Schottkys, suggesting Si more DDD tolerant

• Comparison was of 4H-SiC JBS vs. Si SBD– Si PiN diodes were least DDD tolerant as expected

(minority-carrier device)• Higher-voltage rated devices were less tolerant• At higher temperature, SiC removal rate may be

reduced (see Lebedev, Semiconductors, 2002)


Modified from: Harris, IEEE REDW, 2007

Are Si Schottky’s more DDD tolerant than SiC or is the difference due to device structure and voltage rating differences?

Si vs. SiC

WBG

RH

A

~ 0) G)

:::0 I )>

3.0 ,..... ~ 2.5 <( -

I

I * I

-•- SiC JBS (300V) : -•- SiC JBS (1200V) 1 * -<>- Si SBD (150V)

•

: / -A- Si SBD (60V) 1u 2.0 1 * - • - Si PIN (1600V) & ;I

O - 0 - Si PIN (1200V) •

~ 1.5 *• / / 0 I / ~ • > o •• I 'E 1.0 ;,....:::::...._1-----•--•-•-•-•·• • .,i ~ I • •-•-• 0 .... Ah-• u. 0.5 l-----!----<'.~--~--0~ -v-v..,

t----+---.c.•---.c.~---.c.-A-A-A·A-A I

0.0 ---·=---~~~~------------1 1010 1011 1012

Displacement Damage Dose [MeV/g]

•


• Impact of DDD is temperature dependent:– DDD effects in SiC reduced when measured

at 300°C • In agreement with Lebedev’s 2002 work suggesting

reduced carrier removal rate at higher temperature– Primary effect is donor compensation

• Impacts transconductance, pinch-off voltage, and on-state resistance


Modified from: McGarrity, IEEE TNS, 1992

SiC DDD tolerance increases with temperature, and SiC can operate at these high temperatures

JFET Performance with Dose at 25 °C and 300 °C:

WBG

RH

A

~ OJ G)

::0 I )>

.. -······ .... 0 ••·

0 5

---------------------

10 15 20 Drain Voltage M

50 -.------...,.....~~-.-~~~~~-~ ............. --prerad

< E ....

40

c 30 Cl) ... ... ::::, U 20 C: "§ C

10

- - - 2x1012 MeV/g

•····· 2x1013 MeV/

300 °C

.. ···· .... -······

.. --······· .. • ···················· ....

. ...

0 +-~~~~~-~~-.-~~~~ ~-~........-t 0 5 10 15 20

Drain Voltage M

•


• SiC DDD tolerance depends on polytype– Comparison of 4H- and 6H-SiC JFETs from

Popelka (2014) and McGarrity (1992), respectively, suggests 6H-SiC more tolerant

– Lebedev, JAP 2000, shows that defect behavior differs between 4H- and 6H-SiC:

• Uncompensated donors increase with radiation in 6H-SiC but decrease with 4H-SiC

– McGarrity (1992) calculated a carrier removal rate ~3x lower than Si, suggesting 6H-SiC more DDD tolerant than Si

• Apparent discrepancy with Harris (2007) findings regarding SiC vs. Si may be due to polytype


Modified from: McGarrity, IEEE TNS, 1992; and Popelka, IEEE TNS, 2014

6H-SiC JFET vs. 4H-SiC JFET

For most missions, DDD in SiC will not be a concern. Actual DDD performance depends on device structure, polytype, and application temperature.

Green dashed line: Europa Clipper DDD(Si) behind 100 mil AlFrom: JPL D-80302, Aug 18, 2016.

WBG

RH

A

~ o::J G)

::0 I )> 1.1 ....-----.-................... ,...,.,..._____,----,-......,..,........,-......--.......... ~_ ..... ~~---~-...... _ ...... _ ...... _ ..... ~ ......

a> 1 -IJ- 4H-SiC 9m Uc: I -4- 6H-SiC g m I ____ m

...., 1.0 ;t. - ~ I t---------c---lJ ~4

~ 0.9 \

~ IJ I- 0.8 "C Cl)

-~ m E 0.1 ... 0 z

0.6-+-----.-................... ,...,.,...___.._,----,-......,..,........,-......--.................... ~---.-...................... ...-1

109 1010 1011 1012 1013

~---Displacement Damage Dose [MeV/g]

•

Silicon Carbide:Total Ionizing Dose Effects in MOSFETs

Despite thick oxides, SiC MOSFETs can be TID-robust:

• Cree Gens 1 & 2 in spec up to 100 krad(Si) – Above 300 krad(Si), significant gate-drain capacitance

changes can strongly impact switching performance

• Expect variability between manufacturers and processes– Hole trapping depends strongly on NO anneal time

and temperature– Interface state formation with radiation can vary– Substantial fundamental differences between radiation

responses of Si and SiC MOSFETs found via electrically detected magnetic resonance (EDMR)

TID hardness is coincidental and may change


Plots modified from: Akturk, IEEE TNS, 2012

CRSS

OrbitApprox.

1-year TID[kradISi)]

Europa Clipper 300*

GEO 150

Polar LEO 5

ISS LEO 1*100-mil Ta shielding perJPL D-80302, 8/18/2016

WBG

RH

A

~ OJ G)

::0 I )>

Li:' .s Cl) c.> C _e 0.1 ·u Ill Q. Ill u

~ Cl)

CJ Ill .. 0 > 'ti 0 .c 1/) Cl) .. .c I-

.!: Cl) CJ C Ill .c u

0

-1

-2

-3

-4

-5

-6 0 500 1000 1500

Total Ionizing Dose [krad(Si)]

o prerad a 300 krad(Si) 1::,. 600 krad(Si)

0.01 +---~--~--~-~--~-~ 0 10 20 30 40 50

Drain-Source Voltage [VJ

•

Silicon Carbide Schottky Diodes:Single-Event Effects

• SEEs in SiC Schottky diodes include:– Transient charge collection– Permanent increased leakage current– Catastrophic single-event burnout (SEB)


Kuboyama, IEEE TNS, 2006

Kuboyama, IEEE TNS, 2006

NASA GRC: A. Woodworth, 2015

WBG

RH

A

~ OJ G)

::0 I )>

I I

·.x I

I I

Region 3

F ~- ·1· -------------~::;:~ ~ --~"/-· ;---------------~ 5 I Q)U I ~ I I

I I

--------------------------- T--r---

~ Q) ..., Ol (.) .... Ql ca = .c ou u

Region 1

Bias Voltage [VJ

1 I I I I I I I I I I I I I

•

Silicon Carbide Schottky Diodes:Single-Event Effects – Thresholds

• Onsets for ion-induced leakage current and single-event burnout saturate quickly with linear energy transfer (LET)– Saturation occurs before the high-flux iron knee of the GCR spectrum


Witulski, IEEE TNS, 2018

Typical risk-avoidant mission LET requirements for SEB are > 40 MeV-cm2/mg

WBG

RH

A

~ Cl) G)

::0 I )>

Thresholds 1200 ~~-:;:--:;;-~--.,_ -= -= ~"" ..:..:." :.:....:.~ =.:::.'v':. '..:....:..~ =~ ~~~ =-- ~~~~~~~:::l

1000

-2:. aoo Q)

en 600 ro :!::

g 400

200

00

103..-------:----;------;----,---,--Fcs~ol;:ar:'."-:m~i~ni:m~um:::::===n

102

..... 101

'>. 10° "' "C 10·1

N

·e 10·2

~ 10.J

~ 104

u. 10.s

~ 10-6 Cl

"Iron Knee": max LET(Si) =

29 MeV-cm2/mg

,e 10·7

c: 10-a ~:::._- _----,-ln...,.te_rn_a-::-tio-na-;-l-;;;;Sp::--::a~ce::-;S:;::ta::;;tio::::::-n I- Polar

100-mil Al shielding

10·9 I ~1-==;:::~ ~Ge~o~st~at;::::io9na=ry;::::::;=::::;:::=;:=a.=r~::::--"---:r:::----:~-=-::;:-' 1 o -10 -F

0 1 0 20 30 40 50 60 70 80 90 100

Linear Energy Transfer (Si) [MeV-cm2/mg]

•

Silicon Carbide Schottky Diodes:Single-Event Effects – SEB

• Prior degradation can impact SEB susceptibility, interfering with identification of “SEB safe operating area (SOA)”– At a given reverse-bias voltage (VR), ion-induced leakage current (IR) can impact SEB

susceptibility if SEB does not occur early in the beam exposure• Can’t obtain “highest passing VR” and “lowest VR for SEB” within the same device



Thresholds for SEB have uncertainty:at VR = 485 V, after <104 Xe/cm2, IR = 10 mA with non-linearity of ∆IRbefore that. Is this fluence adequate to rule out SEB? (No)


Adequate heavy-ion test fluence levels are difficult to obtain

WBG

RH

A

• Thresholds 10 1200 - Ion, Bias

1000 - Xe, 475V

~ 8 - Xe, 485V

co -G) > 800 ~ - <(

:::0 QJ E 6 I Ol 600

._..

)> ro .... .:!::::!

C

0 ~ 4 > 400 - ... :::J u

200 2

00 00 50 100 150 200

Time (s)

Silicon Carbide Schottky Diodes:Single-Event Effects – Leakage current

Ion-induced leakage current (IR) findings:• Linearly increases with fluence (independent of flux)• ∆IR depends on

– Applied bias voltage– Ion energy deposition (linear energy transfer (LET))– Ion angle of incidence


Kuboyama, IEEE TNS, 2006 Javanainen, IEEE TNS, 2017

Heat map of ∆IRon log scale: bias and angle dependence with 1217-MeV Xe

Leakage current at 200 V as a function of ion fluence and bias during irradiation

RHA in SiC is challenging: 1) Safe Operating Area to avoid SEB is difficult to identify;2) Non-catastrophic degradation effects on lifetime reliability are unknown

WBG

RH

A

~ CD G)

:::0 I )>

~ .s i: 2! :i ()

<ll C> <1l ~ <1l <ll ...J

al Cl)

<1l 2! (.)

.5

1400

1200

1000

800

600

400

200

0 0 1x104

~ .. • 150V ~

:>

• 145V

e 140V

• 13sv

• 130V

2x104 3x104 4x1a4 5x104 6x1a4

Fluence[particle/dev]

• -340 -7

-320 -8 .:r E

-300 .9 (.)

<t e g ·10 -280 ~ e!

-11 -260 6

'j;j 12 1 -240

~ -13

-220

-14

·200 •40 ·20 0 20 40

anale l°l

Silicon Carbide MOSFETs:Single-Event Effects


SiC MOSFETs exhibit complex effects from heavy ions

• Catastrophic failure or degradation shown in drain and gate currents (ID, IG) during 10 MeV/u Xe irradiation– Results shown are for a 1200-V device

• At all biases above VDS = 50 V, part fails post-irradiation gate stress (PIGS) test – At low VDS, PIGS outcome is dependent on ion fluence

Increasing Drain-Source Voltage (VDS), with Gate-Source Voltage (VGS) = 0V

WBG

RH

A

~ co

~ G')

::0 ... C

I QI ... )>

... :I (J

10·1

10·2

10-3

104

10-S

10-6

10·7

10.a

10·9

10-10

10-11

--Drain Current - - - Gate Current

(abs. value)

10 MeV/u Xe Flux = 20 cm·2s·1

0 VGs• 600 V05

____ ,,,,,---------· --------· \, 10-12 .._ _________ ...,... _____ _

-20 -15 -10 -5 0 5

Elapsed Time [s]

10 MeV/u Xe

-4 Flux = 6 cm·2s·1

~ 2x10 0 VGS• 500 Vos ... C

~ :I

u 1x104

o- - - - ---------------------20 0 20 40 60 80

Elapsed Time [s]

100

1.0x10"6

5.ox10·1

0.0 - --- - ..

-5.0x10·7

.. .. .. .... .... 10 MeV/u Xe

·1,ox10-6 Flux = 1500 cm·2s·1

0 VGs• 300 V05

-1.5x10"6+----------.-------...---...

10-s

~ 10-6 ... C QI 10-7 ... ... :I (J

~ 10.a Ill ~ Ill

10·9 QI ...I QI ... ~ 10-10

-20 0 20 40

Elapsed Time [s]

a Spec/

10 MeV/u Xe: 100 Vos

Slope = 1.3x10·13

Cl a CIC

Cl

a

0

oOO 0

0

10 MeV/u Ar: 300 V05

Slope= 2.6x10·15 10-11-+---~~---~--~----........f

102 103 104 105

Fluence [cm-2]

•

SiC MOSFET Effects as a Function of VDS at VGS = 0 V:Latent Gate Damage

25

Dra

in-S

ourc

e Vo

ltage

Q Collection

During Irradiation Post Run

Measurement Results

No Measurable Effect

IGSS ↑Q Collection

Voltage range of no effects (grey) or only latent damage (change in PIGS test: green) is a function of ion/LET, device voltage rating, and manufacturer/generation

Grey = no ion effects;Green = VDS range for which only latent damage occurs

To be presented by J.-M. Lauenstein at the Radiation Effects on Components and Systems (RADECS) Topical Day, Gothenburg, Sweden, September 16, 2018

WBG

RH

A

1500

~ 10• 1400

co a 0

G) 10• "

_,, 1300

::0 10J 1200

I 10•

)> 10 MeWu Xe: 100 V05

10• S!ope • 13x10'11

a" oOO a" 0

10•ID a 0 10 MeV/u At: 300 VD!

s;-1100 -OJ 1000 tl.O

10'H Slopes:26x 10'tS

10' 10' 10• 10• 10•

ro 900 +-'

Ftuence (cm4) 0 > 800 Q) u 700 lo.. :, 0 600 V')

-----------• I

C 500 ro lo..

0 400

300

200

100

0 Device:

Ion/LET:

I • No Effect

- - - - - -

--

t---

--- --

--

--

--

--

Ml

B 0.9

-

t---

M2 M7

Ar 8.9

-- --

-- --

t---

--

-- --

Ml

Arl0

(to 3300)

Latent Gate Degradation I

-

--

-M2 M3

Cu 23

--

--

--

--

--- -M4 MS

Ag47

--

--

--

--

- - -M2 M3 M4 MG M7 MS

Xe 63

•


26

Dra

in-S

ourc

e Vo

ltage

Q Collection


Measurement Results



No latent damage to gate from low LET/light ions



WBG

RH

A

10•

~ a 0 10• " co

_,, G')

10J

::0 10•

10 MeWu Xe: 100 V05

I 10• S!ope• 13x10'11

)> a" oOO a" 0

10•ID a 0 10 MeV/u At: 300 VD!

10'H Slopes:26x 10'tS

10' 10' 10• 10• 10•

Ftuence (cm4)

-----------•

1500

1400

~ -------------------------- (to 3300)

1300

1200

s;-1100 - ~ ~~~...._

g Q) u lo.. :, 0

V') I

C ro lo..

0

100

0 Device: Ml

Ion/LET: B 0.9

• No Effect

M2 M7

Ar 8.9

Ml

Arl0

Latent Gate Degradation

M2 M3

Cu 23

M4 MS

Ag47

M2 M3 M4 M6 M7 M8

Xe 63

•


27

Dra

in-S

ourc

e Vo

ltage

Q Collection


Measurement Results



Onset is independent of MOSFET voltage rating and manufacturer/generation at higher LETs



WBG

RH

A

10•

a 0 10• "

~

_,, 10J

co 10• G) 10 MeWu Xe: 100 V05

::0 10• S!ope• 13x10'11

a" oOO

I a" 0 10•ID a 0 10 MeV/u At: 300 VD! )> Slopes:26x 10'tS 10'H

10' 10' 10• 10• 10•

Ftuence (cm4)

-----------•

1500

1400

~ -------------------------- (to 3300)

1300

1200

s;-1100 -OJ 1000 tl.O

2 900

g 800 Q)

~ 700 :, o 600

V') I

C 500 ro ,_ o 400

300

200

100

0 Device: Ml

Ion/LET: B 0.9

• No Effect

M2 M7

Ar 8.9

Ml

Arl0

Latent Gate Degradation

•

MOSFET Effects as a Function of VDS at VGS = 0 V:Degradation During Beam Run

28

Dra

in-S

ourc

e Vo

ltage

Q Collection


Measurement Results



Not all MOSFETs exhibit drain-gate leakage current degradation:Design techniques may eliminate this vulnerability

Yellow = VDS range for ∆ID = ∆IG degradationGreen = VDS range for which only latent damage occurs

↑ IDG∆ID = ∆IG

IGSS , IDSS ↑;Failed IGSS

Color gradients span between

known VDSfor given

response types


WBG

RH

A

~ co G')

::0 I )>

·······~ . 1,0X1 0

5.0x10 -1

··•"--- ~

-1 .0xtO,. :~~~~ecn,·~ ·•

.. .. ....

0 Vos, 300 Vos -t.Sxto•+---~-~----

-20 0 20 •• E13psed Time [s)

i -----------•

-----------•

1500

1400

1300

1200

s;1100 -OJ 1000 tl.O ro 900 ...... 0 > 800 Q) u 700 .... :::, 0 600 V)

I

C 500 ro .... 0 400

300

200

100

0 Device :

Ion/LET:

(to 3300)

I • No Effect Latent Gate Degradation D ld=lg I

- - - - - - -

----

--

--

--

--

Ml

B 0.9

f----

-

M2 M7

Ar 8.9

--

--

-- --

-- ---

--- --- --

Ml @ M3

Ar 10 Cu 23

--

--

--

--

f----

--

-

M4 MS

Ag47

--

--

-- --

- f---- -

M2 M3 M4 MG M7®

Xe 63

•

MOSFET Effects as a Function of VDS at VGS = 0 V:Degradation During Beam Run

29

Dra

in-S

ourc

e Vo

ltage

Q Collection


Measurement Results



IDS degradation least influenced by electric field and ion LET: linked to material properties??

Blue = VDS range for ∆ID >> ∆IG degradationYellow = VDS range for ∆ID = ∆IG degradation



↑ IDG, IDS∆ID >> ∆IG

IDSS ↑;↑ or Failed IGSS


known VDSfor given

response types


WBG

RH

A

~ co G)

::0 I )>

:S.0ll 10""' - 10

2.lb:10" •••• la

SOOVir,5 2..0:1:10"' ~ Mev:xe

Ave. Flux: 6 cm4 1-·1

1.&:-10 ...

1.0:ic 10 ...

0.0

... .. Elap$ed Time (•)

i

100

i

-----------•

i i -----------•

-----------•

1500

1400

1300

1200

s;1100 -OJ 1000 tl.O .g! 900

g 800 Q) u 700 .... :::, 0 600 V)

I

C 500 ro .... 0 400

300

200

100

0 Device :

Ion/LET:

~ ------------------------ (to 3300)

• No Effect

M l M 2 M 7

B 0.9 Ar 8.9

La t ent Gat e Degradation • ld=lg • ld>lg

M l M 2 M 3

Arl0 Cu 23

M4 MS

Ag47

M2 M3 M4 M G M7 MS

Xe 63

•

MOSFET Effects as a Function of VDS at VGS = 0 V:SEB/GR

30

SEB/GR vulnerability at LET(Si) < 1 MeV-cm2/mgVulnerability saturates before the GCR flux “iron knee”

Red = VDS range for SEB/GRBlue = VDS range for ∆ID >> ∆IG degradationYellow = VDS range for ∆ID = ∆IG degradation


known VDSfor given

response typesDra

in-S

ourc

e Vo

ltage

Q Collection


Measurement Results





↑ IDG, IDS∆ID >> ∆IG

IDSS ↑;↑ or Failed IGSS

Suddenhigh-I event

Catastrophic Failure:BVDSS < 2 V

SEB

/GR

ons

et n

ot fo

und


WBG

RH

A

-----------•

i i

-----------•

i i -----------•

-----------•

1500

1400

1300

1200

s;1100 -OJ 1000 tl.O ,g! 900

g 800 Q) u 700 .... :::, 0 600 V)

I

C 500 ro .... 0 400

300

200

100

0 Device:

Ion/LET:

~ ------------------------ (to 3300)

Ml

B 0.9

D No Effect Latent Gate Degradation D ld=lg • ld>lg • SEE

M2 M7 Ml

Ar 8.9 Arl0

M2 M3

Cu 23

M4 MS

Ag47

M2 M3 M4 MG M7 MS

Xe 63

•

• Reduced SEB/GR susceptibility– Thicker epilayer

SiC Radiation Hardness by Process: 1200 V MOSFET

31

Just as with silicon MOSFETs, epilayer optimization must be done to balance SEE tolerance with on-state resistance (RDS_ON) performance


known VDSfor given

response types


After Zhu, X., et al., 2017 ICSCRM

WBG

RH

A

• 1500

1400 D No Effect • Latent Gate Degradation • ld=lg • ld>lg • SEE

1300

1200

~1100

Q) 1000 tl.0 ro 900 +-'

g 800 Q) u 700 ,._

~ =, 0 600 co V)

G) I

C 500 :::0 ro ,._ I Cl 400 )>

300

200

100

0 Device: Baseline Modified

Ion/LET: Ag47


• Degradation of IDG eliminated – Drain neck width reduction


32

Elimination of drain-gate leakage current degradation mode is possible through established silicon MOSFET hardening techniques


known VDSfor given

response types



WBG

RH

A

• 1500

1400 D No Effect • Latent Gate Degradation • ld=lg • ld>lg • SEE

1300

1200

~1100

Q) 1000 tl.0 ro 900 +-'

g 800 Q) u 700 ,._ =,

~ 0 600 V)

I

OJ C 500 G) ro ,._ ::0 Cl 400 I

300 )>

200

100

0 Device: Baseline Modified

Ion/LET: Ag47


• Degradation of IDG eliminated – Drain neck width reduction

• Minimal change in onset of other degradation effects:– ∆ID >> ∆IG– latent gate damage

• Rate of degradation at a given voltage is reduced


33

Continued research and development efforts are necessary to understand residual degradation mechanisms!


known VDSfor given

response types



WBG

RH

A

1500

1400

1300

1200

~1100

Q) 1000 tl.0 ro 900 +-'

g 800 Q) u 700 ,._ =, 0 600 V)

I

~ C 500 Cl) ro ,._ G) Cl 400

::0 300 I )> 200

100

0 Device:

Ion/LET:

D No Effect • Latent Gate Degradation • ld=lg • ld>lg • SEE

.....,_._._..,.,,.. ... ,_ -_ -_ -_ -_ -_ -_ -_ -_ -_ ..... ---1

~==========1 • Baseline Modified

Ag47

•

Normalized Onset Voltage for Immediate SEB/GR: Comparison of Device Types

34

Measurable onset for SEB/GR similar across device types: ~40% to 50% of rated VDS

Must derate below this level to account for SEB/GR SOA uncertainty


WBG

RH

A

1.0

0.9

0.8 Q)

g> 0.7 ~ 0 0.6 >

"C 0.5 Q)

.!::! n, E

0.4 '- 0.3 0

~ z co 0.2 G)

:;o I

0.1

• 0.0 0

SEB/GR: 'ff MOSFET 0 Schottky Diode

• JFET ~ PIN Diode

I i

5 10 15 20 25 30 35 40 45 50 55 60 65 70

Linear Energy Transfer (Si) [MeV-cm2/mg]

•

Comparison of SEB/GR Susceptibility Across SiC Power Devices: Neutron Data


The data strongly suggest that for SEB/GR, the electric field strength is driving the susceptibility; there is no indication of a different mechanism

involved for the different device types.

Used with permission from: CoolCAD Electronics (left), Wolfspeed (right)

Terrestrial neutron failures-in-time (FIT) data normalized to the active die area, as a function of bias voltage normalized to measured avalanche voltage.

WBG

RH

A

~ OJ G)

::0 I )>

. . X ........ ...... ....... ; ... ....... ..... ...... ; ...... ,. ... ....... . .

(1) :X •• ,,,,,,, .......... ,,,,!••·················••:- ••·················· ~ 103

cu • .... , (1) • ,, ... .

Cl) . , , .

IA\ 102 , . , . . . , ... , ... . ... . .. j, .. ~ ... ,t,,, .• '. . . . . ; .. ... ... . . . . ....... , v:::!I . , •

N !.a.. • : 5 101

........ . .......... t F, ........ • Wolfspeed 1700 ~ ,,.. .a. !. X STMicro 1200V u. w O Rohm 1200V

1 o0 ........... .. ...... ; .......... • Microsemi

.

0.4

.a. Wolfspeed 1700

0.6 0.8 BiasN

1 (VN)

ava

10000

1000

N 100 E u t:: 10 LL

1

1 0.1

Sea-level Neutron FIT/cm2

0 .3 0 .4 0.5 0.6 0. 7 0.8 0.9 1.0 1.1

Vo/ Vaval

•

RHA Guidance• SEB/GR safe operating area (SOA) is difficult to reliably define

– Susceptibility quickly saturates before the high-flux iron knee of the GCR spectrum• (Unvalidated) failure rate prediction methods developed for Si power devices may

provide an upper bound– Degradation of gate/drain leakage currents hides onset bias for SEB/GR at all but lowest LETs

• For risk-tolerant applications, margin and unpowered redundancy is advised

• Non-catastrophic damage has unknown longer-term effects – Extent of damage is part-to-part variable– Consider application temperature and functional lifetime requirements

• For risk-tolerant applications, margin and unpowered redundancy is advised– Life tests of damaged parts may reveal higher-likelihood failure modes - sample size will limit

discovery of rarer modes • Remember to consider other non-RHA concerns:

– There are no space qualification standards specific for WBG technologies


WBG

RH

A

~ c:o G)

:::0 I )>

ObscrT cd Failun:

=· Const.int (Ran d om)

Failures

•

.. ·· C:ilr'Out

Failures

••••• . I I

· ··· •• J ••••• • ••••••• •• ••• •• ••••• ••• ( ••••• ••••••• •••••• t

Time

SiC RHA Considerations• There are limited efforts underway to develop radiation-hardened SiC power

devices – Some degradation mechanisms may persist despite RHBD efforts– Impact on device long-term reliability must be established

• Radiation hardening comes with a cost– As with Si power MOSFETs, electrical performance will suffer from hardening techniques

• Testing with lighter ions/lower LETs will reveal nuances between designs and aid on-orbit degradation predictions– Responses are saturated at LETs dictated by typical mission destructive-SEE radiation

requirements– LET should be specified in terms of LET(Si) but penetration range must be for SiC

• Characterization data should include identification of voltage conditions at which different effects occur– Richer dataset will include how susceptibility to these effects changes with ion

species/LET/angle/temperature


WBG

RH

A

V ~ OJ C)

:;o I )>

I

•

GALLIUM NITRIDE HIGH ELECTRON MOBILITY TRANSISTOR (HEMT) RADIATION EFFECTS


WBG

RH

A

/' ~ OJ G)

::0 I )>

I

•

GaN HEMT Outline

• Introduction to GaN high electron mobility transistors (HEMTs)

• Effects of radiation:– Displacement damage dose– Total ionizing dose– Single-event effects

• RF GaN HEMTs• Enhancement-mode HEMTs

• Radiation hardness assurance guidance


Images: NASA JPL, courtesy of L. Scheick

SEE failure sites (ex/ right photo) superimposed onto optical image (left) suggest random distribution

WBG

RH

A

V ~ CD G)

:::0 I )>

•

GaN HEMT Introduction: Device Structure

• Two primary flavors of GaN HEMTs: – Depletion mode (normally on)– Enhancement mode (normally off)

• Normally-off GaN is achieved in several ways:– Cascode with a low-voltage normally-off silicon

MOSFET• Penalty to RDS_ON lessens with higher voltage-rated

GaN HEMTs– p-doped GaN gate

• Used in first commercial eGaN devices• Gate built-in voltage exceeds piezoelectric field

– Recessed Schottky gate• Thinned AlGaN barrier beneath gate to reduce

piezoelectric voltage below Schottky built-in voltage– Flourine-implanted AlGaN barrier

• Traps negative charge in the layer


Typical GaN HEMT structure sets up lattice strain and charge polarization between the AlGaN

barrier and GaN layer, creating a 2-dimensional path for electrons to flow

Energy band diagram showing 2-D electron gas formation (yellow)

AlGaN GaNMetalGate

ECEF

EV

Many gate designs: How will these differences affect space qualification standards development for GaN?

Substrate (SiC, Si, …)

AlN (seed layer)AlGaN Buffer

AlGaN BarrierGateSource Drain

GaN

WBG

RH

A

•

GaN HEMT Introduction: Applications

• RF GaN HEMTs– Initially, RF applications dominated due in part

to depletion-mode operation– Designed to work well in linear region for max

gain, min distortion• Characterization includes incident and reflected wave

power measurements as well

• Enhancement mode GaN HEMTs as switches– Development of normally-off HEMTs provided

in-road to power management applications• Start-up does not require an initial gate bias to be applied

to avoid short-circuit damage

– Designed to work well in on-state/transition states to minimize losses

• Low RDS_ON and off-state drain leakage current (IDSS)


Can a single set of radiation test method standards be developed for both RF GaN and enhancement-mode GaN HEMTs?

Output Signal

InputSignal

RF apps often operate in linear region

Switch apps operate along load line

WBG

RH

A

•

Gate-Source Voltage M

::f. c ~ :::,

(.) v C: -~ ~ 0 OJ

G)

::0 I )>

Drain-Source Voltage [VJ

GaN HEMTs:Displacement Damage Dose Effects

• Parametric degradation in GaN HEMTs occurs at DDD levels far above those for space applications– Weaver, et al., 2015 show order-of-magnitude

better performance of GaN vs. GaAs HEMTs– Possibly due to re-injection of

scattered carriers• GaN HEMT DDD effects include

– Decreased drain current– Threshold voltage shift (typically positive)– Decreased mobility and transconductance– Decreased power gain


Weaver et al. ECS J. Solid State Sci. Technol. 2016. Used with permission.

GaN HEMTs out-perform GaAs HEMTs

Ives, IEEE TNS 2015. Used with permission.

Power gain decreases despite correction of ∆ID

For extreme DDD environments, use of a feedback system to maintain quiescent ID will improve

radiation tolerance1.8 MeV protons

WBG

RH

A

✓

~ Cl) C)

::0 I •

1.0

c ~ 0.8 ::, ()

C

-~ 0.6 0

a1 N

~ 0.4 E 0 z

0.2

1E9

! • I "' .., t· Current

••• ~t i a J f Da7

1E10

... . • * ~· "' . • ' "' * t :]

1E11

... * "' X X

• • • • ... ... .

GaAs • GaN

1E12 1E13

Displacement Damage Dose (MeV/g)

13~---------------~

12.5

12

iii' 11 .5 :!:!. c: 11 ·;;; C> 10.5

10

9.5

• Pre

• 3x1012

1x1013

.. 3x1013

• 1x1014

?is -20 -1s -10 -s o s Input Power [dBm]

10 15

•

GaN HEMTs:Displacement Damage Dose Effects

• DDD susceptibility is greater when:– parts are biased during irradiation– parts have had prior hot-carrier stress– see Chen, et al., IEEE TNS 2015

• Radiation increases the concentration of existing defects


Pearton, ECS J. Solid State Sci. Technol. 2016. Used with permission.

Concentrations of existing defects increase with radiation exposure

GaN HEMT DDD effects: 1) Bias matters during irradiation

2) Prior hot-carrier stress can enhance radiation effects3) There is significant variability of response to radiation

WBG

RH

A

~-0.13ev, E -0.16eV-. Ec-0.60eV

E -0.71eV

n-GaN:Si p-GaN:Mg

+3.28eV

----Ev+3.00eV

Ey+2.42eV

---- - Ev+1 .50eV

E -3.25eV ~===~~---- EF VB

/' ~ OJ G')

::0 I )>

as-grown traps irradiation-induced traps

•

• Many gate designs:• RF GaN HEMTs (Schottky gate)

– Aktas, 2004 demonstrated 0.1 V threshold shift after 6 Mrad(Si) γ irradiation

– Harris, 2011 demonstrated no significant shift after 15 Mrad(Si) proton irradiation

• p-GaN gate– 500 krad(Si) γ irradiation: < 18% Vth shift (Lidow,

2011)• MOS gate

– No data– This design is most likely to be sensitive to TID

effects


GaN HEMTs:Total Ionizing Dose

Images: NASA JPL

GaN HEMT designs vary

GaN HEMT TID effects: Robust in absence of gate oxide;

To be determined for those with oxides

WBG

RH

A

V ~ OJ G)

::0 I )>

1 µm >-< EHT=1000kV

WD : 70mm

EHT • 10.00 kV

WO• 71mm

Signal A ., SESI

Mag= t0OOKX

Signal A • SESl

Mag • 3.15KX

Date .28 Apr 2016 FIB Loc:t< Mag$ ., No

o.te .2GA9{2018 F18 Lode Mags ,. No

•

111

• Catastrophic SEE:– In some parts, SEB occurs only at VDS levels above those reached during RF operation

• Rostewitz, 2013 and Osawa, 2017 tested in both DC and RF modes: Calibrated measurements of max transient VDS during RF operation showed no excursions above SEB failure threshold VDS found in DC mode tests

• Osawa, 2017 noted the failure threshold in DC mode was over 3X VDS bias level for RF mode; no failures in RF mode

• Armstrong, 2015 showed no failures in RF mode– However, some unpublished data suggest some parts/lots may be susceptible to SEB in

RF mode• Lot-lot variability found


RF GaN HEMTs:Single-Event Effects

Although generally SEB-robust, some unpublished SEB failures have occurred at biases relevant to RF operation

WBG

RH

A

/ ~ OJ G)

:::c I )>

•

• Increased leakage currents– Armstrong, 2015 study of 5 different HEMTs from 4 manufacturers revealed manufacturer and

part-part variability in susceptibility to increasing gate leakage current due to heavy-ion induced damage

– Kuboyama, 2011 study reveals two different degradation modes• At lower DC VDS bias, IDG degradation• At higher DC VDS bias, IDS degradation


RF GaN HEMTs:Single-Event Effects

All heavy-ion studies of RF GaN HEMTs report increased leakage current

ID and IG as a function of 3.5 MeV/u Kr fluence and VDSreveals 2 damage pathways

Kuboyama, 2011 IEEE TNS. Used with permission.

WBG

RH

A

/ ~ OJ G)

:::0 I ):>

lx l0-31-------------------L.;14*015~ V 8x l04

~ i: ~ 6x l04

u ~ ~ i'3 4xJ0-4 ....J

2x l04

30 40 50

0 5xl07

I Drain Current I J I~

130

90 120

100

11 0 )t:,;jent I

lx l08 Fluence [cm-2]

l.5x I os

•

• Catastrophic SEE:– Susceptibility is very wafer-lot specific

• Likely accounts for some data discrepencies in the literature• Lot-specific testing is critical

– Susceptibility can depend on test circuit design (Scheick, 2014, 2015, and 2018)• When gate is tied to the source at the part level, threshold for SEE improves

– Gate transients upon ion strikes are unrealistically dampened• Gate off-state bias level has minimal effect

– Angle effects are variable• Variability between lots of same part (Scheick, 2018)• Compounds difficulty of estimating an on-orbit failure rate

• Increased leakage currents:– Drain-source leakage current affected


Enhancement-Mode GaN HEMTs:Single-Event Effects

Lot-specific testing is critical

0 20 40 60

Angle [Deg]

0

200

400

600

800

Dra

in-to

-Sou

rce

(Vds

) [V

]

@ g

NASA, courtesy L. Scheick.

Variable effect of angle on SEB susceptibility

WBG

RH

A

~ OJ G)

::0 I )>

•

RHA Guidance• Lot-specific heavy-ion testing is necessary

– Several suppliers now offer “space qualified” enhancement-mode GaN HEMTs, providing wafer lot level screening for SEE• Radiation test methods for GaN are not standardized

• Unlike for SiC, GaN SEE safe operating areas (SOAs) can be identified– Valid for the tested lot only– Part-part variability reduces confidence

• Derate, derate, derate– Ensure that all transient voltages fall within the SOA

• More data are needed on angle effects to understand worst-case test conditions– SEB SOAs to date are established with data taken at normal incidence only

• Until effects are understood, additional margin (derating) may reduce risk

• Effect of leakage current degradation on device reliability/lifetime is unknown• Remember to consider other non-RHA reliability concerns

– Hot carrier damage– Current collapse– Dynamic RDS_on

– And others48To be presented by J.-M. Lauenstein at the Radiation Effects on Components and Systems (RADECS) Topical Day, Gothenburg, Sweden, September 16, 2018

WBG

RH

A

/ ~ OJ G)

::0 I )>

•

A GLIMPSE INTO THE FUTURE


The Future •

Qualification Standardization & Technology Maturation

• Supplier qualification data are based on standards for silicon– Methods vary between vendors

• JEDEC has established a new committee to address this need– JC-70 Wide Bandgap Power Electronic Conversion Semiconductors

• Sub-committees for SiC and GaN– Focused initially on non-radiation reliability

• Aerospace Corporation Guideline document for GaN now available– TOR-2018-00691, “Guidelines for Space Qualification of GaN HEMT Technologies”

• NASA Electronic Parts and Packaging Program RHA guidelines to be developed– Work to begin FY2019

• Funded efforts to mature WBG technologies for space application are underway– ESA Materials and Components Technology Section (TEC-QTC) programs– NASA and Dept. of Defense Small Business Innovative Research (SBIR) awards and other

grant mechanisms– Other international government and industry programs


The Future

•

The Future of SiC and GaN

Markets:• Automotive industry forecasted to drive the SiC power market

– Almost all carmakers are working to implement SiC in their main inverter– SiC transistor forecasted compound annual growth rate for 2017-2023: 50%– Source: Yole Power SiC 2018 Market & Technology Report

• Largest growth in GaN market size will be power supply applications– Over 2 orders of magnitude growth expected by 2021

• Source: Yole Power GaN 2017: Epitaxy, Devices, Applications, and Technology Trends

R&D efforts• Vertical GaN power devices for higher current and voltage• Increased GaN reliability• Increased SiC reliability


Creative Commons CC0

The Future

•

Ultra WBG: Diamond

• Benefits:– Best of everything (see plot)– High electron and hole mobilities– Electron emissivity

• Target applications: “Ultra-high power” market (1 kW or more)– Radar– Communication satellites– Vacuum electronics (driver, cathode)

• Status:– 10 mm x 10 mm wafers commercially available– Vertical SBDs, MESFETs, and MOSFETs have been

developed• Challenges for maturation

– Crystal quality and size– Absence of n-doping


The Future


[x107 emfs]


\ \

\

,'

' \ \

\

\

\


• Bandgap Energy ~ -

[eV] ~ ~~k

,'


~~- 0/4 C' --~

<$>~. ~$ 0~

,S'


,' - Si --4H-SiC --GaN --c


Ultra WBG: Ga2O3

• Benefits:– Power-Frequency figure of merit (FOM) > 2x GaN– Specific on-resistance in vertical drift region > 10x SiC– High-quality native substrate– Wafer costs are comparable to GaN

• Target applications: “Ultra-high power” market (1 kW or more)– Radar– Communication satellites– LED market

• Status:– Least mature of the ultra-WBG technologies– 2” Wafers commercially available– Vertical SBDs, MESFETs, and MOSFETs have been developed

• Challenges for maturation– Crystal quality– Absence of p-doping; self-trapping of holes– High contact resistance– Thermal management: poor heat dissipation


The Future


[x107 emfs]



• Bandgap Energy ~ -

[eV] ~ ~~k


~~- 0/4 C' --~

<$>~. ~$ 0~

,S'


- Si --4H-SiC --GaN --f3-Ga20 3



Thank you for your attention!

•

wide-bandgap semiconductors in space: appreciating the

Documents